I was experimenting for the last 2 hours attempting to create a more bloodbound-oriented gem that would eventually be better than the current 32-spec gem. After at least 6 ratio failures and many more supergem failures, I stumbled on this jackpot.

This is the Standard 32-spec triple gem-

This is my Gem of Subtle Fury-

Looks inferior? It is! For a while anyway.

Here are the gems I used to 1:63 both to squish the red more(gaining a universal 3.7% mana leech). I replaced the red in both formulas with orange.

The 32-spec-

The Gem of Subtle Fury-

After two iterations of the 64-combine, here is the 32-specced gem with 400 hits-

Here is my Gem of Subtle Fury-

Same hit level, just pointing that out.

Now, the kicker. In a trap with G16 64-combined pure O amps...

The 32 spec-

Aaaaaaand....the Gem of Subtle Fury(dang I like this name!) -

Only a 0.2% difference....but the GoSF was inferior with both having no hits. Which means it gets better the higher they both go in hit levels. Again, its hardly anything, but it IS better. Numbers don't lie.

Here's the horrid Paint drawing of the formula. The ONLY things I changed were Red's position and introducing black to the orange earlier. Incredible, huh?

If it's too confusing, basically this is how I built the top portion(19 G1s equivalent), the bottom is identical.

Create G1 Orange
Create G1 Red
Combine
Create G1 Black
Combine G2 RO into B
Create G2 Orange and 2 G1 Orange
Combine the G2 O with both G1s
Combine the Triple onto that.
Create Dual Orange Black G2
Create G1 Black
Combine Dual with G1 Black
Combine Triple with Dual.
Create G2 Orange and G1 Black
Combine Orange onto Black
Combine Triple with the new Dual.
Create G2 O, G1 O, G1 B, G2 B.
Combine G2 O onto G1 O, G1 B, G2 B in that order.
Combine result with Triple.

I found one even better than my GoSF. It is now named the replica of the GoSF and the new better one is the GoSF.

The old GoSF's scheme.

o = orange
b = black
r = red

Ratio: 1 red / 14 orange / 17 black

A = (o + r + b) + (((o + o) + o) + o)
B = (A + ((o + b) + b)
C = (B + ((o + o) + b)
D = ((((o + o) + o) + b) + (b + b))
E = Duplicate D
F = C + D1
G = ((((b + b) + b) + b) + b) + (b + b)
H = G + D2
I = F + H
J = Complete.

The New and Improved(+0.3% more leech at Grade 20 supergem, 6 amp G21, zero hits) Gem of Subtle Fury-

Ratio: 1 red / 12 orange / 19 black

A = (o + r + b) + (((o + o) + o) + o)
B = A + ((o + o) + b)
C = B + ((o + b) + b)
D = (((o + b) + o) + b) + (b + b)
E = Duplicate D
F = C + D1
G = ((((b + b) + b) + b) + b) + (b + b)
H = G + D2
I = F + H
J = Complete

The result at grade 5 = +0.02 chain, -0.68 leech, + 0.005 per hit level.

in the end, bloodbound oriented gems beat all.

And now I shall start my mana run. on Cache GCCS with a safe-ish 4995 monster.

The benefit is pretty decent, at G72 equivalent it is 1.5% better.

Not much, but multiply all the factors.

16 combine is about 4x better at this point(combine with 32-spec) 64 combine is 25% better than that. Squishing Red once is 3.7% better than that. New combine is 1.5% better than that.

4.00 * 1.25 * 1.037 * 1.015 = 5.26x better than a U'd gem.

4.00 * 1.25 = 5

5 * 1.037 = 5.18

5.18 * 1.015 = 5.26.

So overall it gives an added benefit of +8% better than U-upgrading.

This kind of begs the question : for a given orange:black ratio, at what hit level does the increase in black start to compensate for the decrease in orange?

It also begs the followup : at what hit level does an orange:black ratio of 1:30 (leaving that last slot for red) become the best ratio of all?

And the follow-up to the follow-up : would it be better to "squeeze" the ORANGE out, relying solely upon amplifiers, and just have the black boost the gem?

At grade 5 the new one is better by 2.06 mana leech. Clearly, bloodbound oriented gems gain power later on. This gem is worse than the old 32-spec.

And if it is worse than the 32-spec, it is MUCH worse than my special combines. Haha Psorek lol your calculation machine failed it seems. Unless I misunderstood it?

My Grade 5 stats maxed skills out of trap-

32-spec

4.85 chain 8.85 leech +0.254/hit level.

New one

4.57 chain 9.88 leech +0.239/hit level

In trap with amp-

Leech 285 VS Leech 287.06

However, clearly by grade 15 supergem, the new one is inferior. And again, if it's inferior to the old 32-spec, that means my gem of subtle fury is even better than it.

Hehe. Looks like I'm 2-0 for you Psorek! Sorry lol

Oh, and I think that the problem with it is that frankly, for both the 6 - grade 1 - value - grade 3's,

the optimal, the MOST optimal combination for maximum black with high and un-squished orange, is the way psorek did it in the old combine- G2 O, G1 O, G1 B, G2 B.

The reason being, the first combination strengthens the orange component. The second introduces a relatively weak black component, allowing the combine to throw away stats from the black grade 1 instead of from the supergem orange. This is why supergemming works btw. It's how combining works.

So, after the 2nd combine, you get a strong-orange with a weak-enough-in-order-to-save-the-orange-part black, but enough black to make it a strong trait. The third combine strengthens the black vastly, while minimizing the orange's loss, as combining a gem worth 4 G1 with one worth 2, again the black's stats, EXCEPT specials, are thrown mostly away. And the part from the 4-g1 grade 2 thrown is the less-recent stats, eg, the orange. But it's minimal, which results in a strong-black-medium-orange gem.

The best combination, anyway. Doing black then orange results in a much weaker black component, and only slightly more orange. Don't ask me why lol.

But basically look at the stats I'll show below comparing the 6-gem G3 from 32-spec and the new one.

It also goes against the GCL-GC2 supergemming laws, in which the closer the grades, the better overall the supergemming will be in aiding the supergemmed gem. See, a G1 into a G3 will only mildly influence the G3, as said G3 already has 5 components(5 G1s) that a sixth will only mildly influence it. Add to the fact that it is a black G1, a weak component when combining(YES it is weak in the fact of how much power it gains between levels, white is also weak. I believe it gains +9% per level?)

When combining, it dilutes the black obviously, and since black is already harder to supergem(supergemming effectiveness dies out around the third pure black ONE grade lower added, while for orange it can withstand up to 5(that's why we see it still be effective by then, while each black G1 only adds less than +0.01 per hit level at the low grades.

Combining it into an orange-oriented 2 grade higher gem which already was supergemmed once, will hardly effect it and give a lesser effect than the old combine.

At least, that is my theory and seems to fit for a lot of these formulas.

Now the stats will prove my theory right-

Old 32-spec

G2 - G1 - G1 - G2 = G3(G2 is supergemmed by G1 = most effective)

4.66 leech +0.281/hit level.

New 32-spec, SUPPOSEDLY better(BTW, I am the first to actually TEST to see if it's actually better? Huh!)

5.97 leech +0.248/hit level.

-------

Now, Astro suggested something.

The reason this is even LESS effective than the above is because of a main rule in combining: When you combine two pure gems into a dual for the first time, some extra percentage of specials from BOTH sides are discarded beyond the base special penalty of in this case, 25% reduction(75% power for both)

Take my example-

G1 Orange + G1 Yellow

G1 orange = 3.71 per hit.

G1 Yellow = 80% to hit +333% harder. Why do I say this way instead of x4.33? Simple: The base yellow HAS to be 1x. You can't hit less than 1.00x damage. Thus, the actual special bonus atop that 1x is indeed +3.33x, or +333%

Now, when combined....

G2 Orange-Yellow

2.45 Mana Per Hit(66.03% of the grade 1)

66.03% / 75%(what it should be) = 0.88

In other words, besides the dual gem penalty, you lose an additional 12% mana leech. Thus, combining a once-combined dual into a supergem is a big no-no, as the dual has reduced stats than what it should be in BOTH specials.

For the critical hit,

Dual - 64% to deal +2.2x damage.

Now first, we need to see that 20% is already taken off.

Next, 2.2 / 3.33 = 66.06%! What a surprise, exactly/almost the same penalty as the leech!

66.06 * 0.8 = 52% as much power as the G1! Because it deals crits 20% less frequently as well!

So you see, immediately combining a new-dual into a supergem lowers everything from what it could be. I found this out due to my experimenting and that's EXACTLY why the FIRST combination in my version of this part of the 32-spec, has the dual gem as the VERY FIRST combination. The loss in specials is recuperated by repeated combining of both the gem's specials as supergemming.

So thus, that is why the formula is currently bad. The old 2-1-1-2 is the very best except for my formula, which uses dual-1-1-2, which is slightly more effective...with amps.

Huge post on Gem Lore, but I had to share my theories suggestions and criticism somewhere :P Hope it makes some sense.